U.S. patent application number 13/866302 was filed with the patent office on 2013-10-24 for magnetic resonance method and apparatus for obtaining a set of measured data relating to a breathing object of interest.
The applicant listed for this patent is Alto Stemmer. Invention is credited to Alto Stemmer.
Application Number | 20130281823 13/866302 |
Document ID | / |
Family ID | 49290177 |
Filed Date | 2013-10-24 |
United States Patent
Application |
20130281823 |
Kind Code |
A1 |
Stemmer; Alto |
October 24, 2013 |
MAGNETIC RESONANCE METHOD AND APPARATUS FOR OBTAINING A SET OF
MEASURED DATA RELATING TO A BREATHING OBJECT OF INTEREST
Abstract
In a method and apparatus to acquire a measurement data set of a
breathing examination subject by magnetic resonance, the
measurement data set is acquired in multiple shots each composed of
a number of k-space trajectories (views), with the number Nv of
views per shot being selected. The number of shots is determined in
order to completely fill k-space to be scanned. The views of the
shots are associated with sectors in k-space, such that
approximately the same number of views are arranged in each sector,
and such that all views in a sector have a similar distance from
the k-space center. A respective view of each sector is associated
with a respective one of the shots, corresponding to the
orientation of the respective shot in the kz-ky plane. The views of
each shot are scanned such that views that are associated with the
same sector and different shots respectively assume the same time
position within the shot.
Inventors: |
Stemmer; Alto; (Erlangen,
DE) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
Stemmer; Alto |
Erlangen |
|
DE |
|
|
Family ID: |
49290177 |
Appl. No.: |
13/866302 |
Filed: |
April 19, 2013 |
Current U.S.
Class: |
600/410 |
Current CPC
Class: |
A61B 5/0037 20130101;
A61B 5/113 20130101; A61B 5/055 20130101; G01R 33/5676 20130101;
A61B 5/7207 20130101; G01R 33/4818 20130101 |
Class at
Publication: |
600/410 |
International
Class: |
A61B 5/055 20060101
A61B005/055 |
Foreign Application Data
Date |
Code |
Application Number |
Apr 20, 2012 |
DE |
102012206547.4 |
Claims
1. A method to acquire a magnetic resonance data set of a breathing
subject, comprising: operating a magnetic resonance data
acquisition unit to obtain a magnetic resonance data set by
acquiring magnetic resonance data from a breathing subject in
multiple shots each shot comprising at least one view composed of
k-space trajectories along which magnetic resonance data of the
respective shot are entered into k-space; in the processor,
selecting a number Nv of years per shot; in said processor,
determining a number Ns of shots required to completely fill
k-space; in said processor, associating the years of the Ns shots
with Nv sectors in k-space to cause substantially a same number of
years to be arranged in each sector, and to cause all views in a
sector to have a substantially similar distance from a center of
k-space; in said processor, associating a respective view of each
sector with a respective one of the Ns shots, dependent on an
orientation in a kz-ky plane in k-space; and from said processor,
entering magnetic resonance data for the views of each shot into
k-space with views associated with a same sector and different
shots respectively having a same time position within the respected
shots.
2. A method as claimed in claim 1 comprising associating said
respective years of each sector to cause years associated with a
sector to be within a same half of said kz-ky plane.
3. A method as claimed in claim 1 comprising associating with years
of a respective sector with a respective one of the Ns shots
according to an azimuthal angle in a polar coordinate system in the
kz-ky plane.
4. A method as claimed in claim 1 comprising entering the magnetic
resonance data for the respective views of each shot into k-space
in an order corresponding to a neighborhood of the sectors.
5. A method as claimed in claim 1 comprising: acquiring a current
respiratory signal with a sensor from the subject before acquiring
magnetic resonance data of a respective shot; in said processor,
assigning a shot index; in said processor, assigning a shot index
.epsilon.=[0; . . . ; ns0; . . . ; Ns-1] to each shot, which shot
index describes a neighborhood of the shots in k-space, wherein the
shot indices ns are ordered such that the sensitivity of the shots
with regard to a movement of the examination subject increases from
the shot index ns=0 to the shot index ns=ns0 and decreases from the
shot index ns=ns0 to the shot index ns=Ns-1; in said processor,
associating a shot to be measured after a measurement of the
breathing signal with a cluster, corresponding to the breathing
signal measured with the navigator measurement, wherein a breathing
signal range is unambiguously associated with a cluster, and a
cluster encompasses all shot indices of shots that have already
been acquired after the measurement of a breathing signal in the
breathing signal range of the cluster; in said processor, selecting
a shot index for the shot to be measured depending on the shot
indices already acquired by the previously selected cluster and its
immediately neighboring clusters, and acquire the shots belonging
to this shot index after the measurement of the breathing signal,
wherein the neighborhood of clusters is defined depending on the
breathing signal range associated therewith; operating said
magnetic resonance data acquisition unit to acquire a respective
shot after the navigator measurement that has a shot index selected
depending on the shot indices already acquired in the cluster with
which the respective shot is associated; and generating said
magnetic resonance data acquisition unit to acquire navigator
measurements and shots until a predetermined number n of adjacent
clusters includes common shots with all Ns shot indices.
6. A method as claimed in claim 5 comprising, in said processor,
selecting a shot index for a shot to be acquired by selecting one
of n cluster combinations that respectively comprise the cluster
associated with the shot to be acquired, and n-1 clusters adjacent
to said one of said n cluster combinations, said one of said n
cluster combinations being a cluster combination in which a largest
number of shots with different shot indices have already been
acquired.
7. A method as claimed in claim 6 comprising associating a shot
index with each cluster as a start position for acquisition of
magnetic resonance data in the shot, and selecting a shot index for
a shot associated with a cluster in which no shot has previously
been acquired.
8. A method as claimed in claim 6 comprising increasing a shot
index by one relative to a previously highest shot index, and
selecting said shot index increase by one for a shot that is
associated with a cluster having a start position in a region of a
left k-space edge with low sensitivity to said movement and with
which at least one preceding shot was already associated, and a
shot decreased by one relative to a previously lowest shot index is
selected for a shot that is associated with a cluster with which a
peripheral start position at a region of a right edge of k-space
with a low sensitivity to said movement, and with which at least
one preceding shot was already associated.
9. A method as claimed in claim 6 comprising selecting a shot index
increased by one with respect to a previously highest shot index
for a shot that is associated with a cluster having a central start
position in a region with high sensitivity to said movement and
with which at least one preceding shot was already associated when,
in the selected cluster combination, a set of shot indices that had
not yet been acquired in the selected cluster combination is
greater in the shot index range ns.gtoreq.ss0 than in the shot
index range ns.ltoreq.ns0, and selecting a shot index decrease by
one relative to a previously lowest shot index for a shot that is
assigned to a cluster having a central start position in a region
with high sensitivity to said movement and with which at least one
preceding shot was already associated when, in the selected cluster
combination, a set of shot indices that had not yet been acquired
in the selected cluster combination is greater in the shot index
range ns.ltoreq.ns0 than in the shot index range ns.gtoreq.ns0.
10. A method as claimed in claim 5 comprising assigning a central
start region in an index range with a high sensitivity to said
movement and a peripheral start position at an edge of the index
range to adjacent cluster, and selecting said start position as a
shot index for the shot to be acquired as long as no shot has
previously been acquired for the cluster with which the shot to be
acquired is associated.
11. Method according to claim 10, wherein selecting a shot index
for a shot to be measured comprises selection of one of the two
possible cluster combinations that respectively include the cluster
with which the shot to be measured has been associated and one of
the two clusters immediately adjacent thereto, wherein that one of
these cluster combinations is selected in which the most shots with
different shot indices have already been acquired.
12. Method according to claim 11, wherein the selection of a shot
index for a shot to be measured that has been associated with a
cluster to which the peripheral start position has been assigned
takes place such that the lowest shot index that has not yet been
acquired by the selected cluster combination is selected in the
event that the set of shots with shot index ns.ltoreq.ns0 that have
not yet been acquired in the selected cluster combination has more
elements than the set of shots with shot index ns.gtoreq.ns0 that
have not yet been acquired in this cluster combination; and the
highest shot index that has not yet been acquired by the selected
cluster combination is selected in the event that the set of shots
with shot index ns.gtoreq.ns0 that have not yet been acquired in
the selected cluster combination has more elements than the set of
shots with shot index ns.ltoreq.ns0 that have not yet been acquired
in this cluster combination.
13. Method according to claim 11, wherein selecting a shot index
for a shot to be measured that has been associated with a cluster
to which the central start position has been assigned is
implemented, as long as at least one preceding shot has already
been associated with this cluster, by selecting a shot index that
has been increased by one relative to the previous highest shot
index encompassed by the central cluster for the shot to be
measured in the event that the set of shots with shot index
ns.gtoreq.ns0 that have not yet been acquired in the selected
cluster combination has more elements than the set of shots with
shot index ns.gtoreq.ns0 that have not yet been acquired in this
cluster combination; and a shot index that has been decreased by
one relative to the previous lowest shot index encompassed by the
central cluster is selected for the shot to be measured in the
event that the set of shot indices [sic] with shot index
ns.ltoreq.ns0 that have not yet been acquired in the selected
cluster combination has more elements than the set of shots with
shot index ns.gtoreq.ns0 that have not yet been acquired in this
cluster combination.
14. A magnetic resonance apparatus comprising: a magnetic resonance
data acquisition unit; a processor configured to operate said
magnetic resonance data acquisition unit to obtain a magnetic
resonance data set by acquiring magnetic resonance data from a
breathing subject in multiple shots each shot comprising at least
one view composed of k-space trajectories along which magnetic
resonance data of the respective shot are entered into k-space;
said processor being configured to select a number Nv of years per
shot; said processor being configured to determine a number Ns of
shots required to completely fill k-space; said processor being
configured to associate the years of the Ns shots with Nv sectors
in k-space to cause substantially a same number of years to be
arranged in each sector, and to cause all views in a sector to have
a substantially similar distance from a center of k-space; said
processor being configured to associate a respective view of each
sector with a respective one of the Ns shots, dependent on an
orientation in a kz-ky plane in k-space; and said processor being
configured to enter magnetic resonance data for the views of each
shot into k-space with views associated with a same sector and
different shots respectively having a same time position within the
respected shots.
15. A non-transitory, computer-readable data storage medium encoded
with programming instructions, said data storage medium being
loaded into a computerized control and processing system of a
magnetic resonance apparatus, said magnetic resonance apparatus
comprising a magnetic resonance data acquisition unit, and said
programming instructions causing said computerized control and
processing system to: operate a magnetic resonance data acquisition
unit to obtain a magnetic resonance data set by acquiring magnetic
resonance data from a breathing subject in multiple shots each shot
comprising at least one view composed of k-space trajectories along
which magnetic resonance data of the respective shot are entered
into k-space; select a number Nv of years per shot; determine a
number Ns of shots required to completely fill k-space; associate
the years of the Ns shots with Nv sectors in k-space to cause
substantially a same number of years to be arranged in each sector,
and to cause all views in a sector to have a substantially similar
distance from a center of k-space; associate a respective view of
each sector with a respective one of the Ns shots, dependent on an
orientation in a kz-ky plane in k-space; and enter magnetic
resonance data for the views of each shot into k-space with views
associated with a same sector and different shots respectively
having a same time position within the respected shots.
Description
BACKGROUND OF THE INVENTION
[0001] 1. Field of the Invention
[0002] The invention concerns a method to acquire a measurement
data set of a breathing examination subject by means of magnetic
resonance technology; a magnetic resonance system; a computer
program; and an electronically readable data medium.
[0003] 2. Description of the Prior Art
[0004] Magnetic resonance (MR) is a known technology with which
images from the interior of an examination subject can be
generated. Expressed simply, the examination subject is placed in a
magnetic resonance imaging scanner, in a strong, static, homogenous
base magnetic field, also called a B.sub.0 field, having a field
strength of 0.2 tesla-7 tesla and more, such that the nuclear spins
of the subject orient themselves along the base magnetic field. In
order to trigger magnetic resonance signals, the examination
subject is irradiated with high frequency excitation pulses (RF
pulses), the triggered magnetic resonance are detected and entered
into a memory that represents a mathematical domain known as
k-space, and MR images are reconstructed on the basis of the
k-space data, or spectroscopy data are determined. For the spatial
encoding of the measurement data, rapidly activated magnetic
gradient fields are superimposed on the base magnetic field. The
recorded measurement data are digitized and stored as complex
number values in a k-space matrix. From the k-space matrix
populated with data values in this manner, an associated MR image
can be reconstructed, for example, by means of a multi-dimensional
Fourier transformation.
[0005] The respiratory movement of a patient that is to be examined
by means of MR can lead to so-called ghosting, to blurring, and/or
to intensity losses in the images generated, as well as
registration errors between generated images particularly in an
examination of the organs of the thorax and the abdomen, i.e. of
examination regions affected by respiratory movement. These
artifacts can make it difficult for a physician to perform an
analysis on the basis of the images, and can lead to lesions being
overlooked, for example. Numerous techniques exist in the prior art
for reducing artifacts resulting from respiratory movement. One of
these techniques is respiratory gating. Respiratory gating is a
technique with which, during the MR measurement, the respiration of
the patient is recorded and assigned to the acquired measurement
data. With respiratory gating, only measurement data are then used
for reconstruction for which the associated recorded respiratory
movement fulfills certain specifiable criteria.
[0006] The breathing of the patient can be detected with external
sensors, for example a pneumatic cushion, or with MR signals (known
as navigators). A navigator is normally a short sequence that
acquires MR signals, for example of the diaphragm or another signal
source in the examination subject whose movement is correlated with
the breathing of the patient. The breathing movement can be
reconstructed via the position of the diaphragm or the other signal
source.
[0007] In breath gating with navigators, the navigator sequence is
(for example) interleaved with the imaging sequence, and a
diaphragm position measured with a navigator is subsequently
associated with the imaging data acquired immediately following (or
before) this.
[0008] A distinction is made between retrospective and prospective
respiratory gating.
[0009] With retrospective respiratory gating the respiratory
movement is detected and recorded during the MR measurement, but
not evaluated. Instead, the k-space that is to be recorded is
measured repeatedly. For the reconstruction, only a portion of the
measured data are referenced, preferably that data in which the
respiratory signal lies within a specific window for a distinctive
respiratory position. If a specific k-space data point that is
necessary for the image reconstruction is repeatedly measured
within the distinctive window, then the data can be averaged. If,
instead, a data point is always measured outside of the window,
then that data point deviating the least from the distinctive
position can be used for the reconstruction.
[0010] With prospective respiratory gating, the physiological
respiratory signal measured using a respiratory sensor (e.g. the
diaphragm position measured with a navigator sequence) is evaluated
during the measurement, and the MR measurement is controlled, based
on the recorded physiological signal. In the simplest embodiment,
the so-called acceptance/rejection algorithm (ARA), the measurement
of an imaging data packet (and if applicable, the associated
navigator sequence) is repeated until the physiological signal
falls within a previously defined acceptance window.
[0011] One example of an acceptance/rejection algorithm of this
type and, at the same time, the first description of respiratory
gating with navigators, is described in the article by Todd S.
Sachs, Craig H. Meyer, Bob S. Hu, Jim Kohli, Dwight G. Nishimura
and Albert Macovski: "Real-Time Motion Detection in Spiral MRI
Using Navigators," MRM 32: Pages 639-645 (1994). The authors
acquired one or more navigators for each excitation of a spiral
sequence. The navigators were acquired here following the
acquisition of the image data. Different navigators are
distinguished by their spatial orientation. From each navigator, a
spatial displacement along the axis of the navigator in relation to
a reference navigator is calculated using a cross-correlation. The
navigator scan acquired following the first imaging scan is used,
in each case, as a reference. A specific imaging scan is repeated
until the spatial displacement determined with the navigator, in
relation to the reference, is less than a threshold value provided
by a user. This, therefore, relates to an acceptance/rejection
algorithm based on one or more spatial displacements.
[0012] Another example of an acceptance/rejection algorithm is
described by Wang et al. in "Navigator-Echo-Based Real-Time
Respiratory Gating and Triggering for Reduction of Respiratory
Effects in Three-Dimensional Coronary MR Angiography," Radiology
198; Pages 55-60 (1996). In this case, the physiological signal is
the displacement of the diaphragm position, determined with a
navigator, in relation to a reference state. One difference from
the work by Sachs et al. is that, in each case, a navigator is
acquired before and after the imaging scan, and that the imaging
scan is then only accepted if the displacement determined by means
of both navigators is less than the threshold value.
[0013] In order to determine the acceptance window, a so-called
pre-scan is normally carried out for each patient, in which the
respiratory movement is recorded, for example, with the navigator
sequence, but imaging data are not yet acquired.
[0014] Prospective respiratory gating is normally more efficient
than retrospective respiratory gating. A prerequisite for
prospective respiratory gating is a real-time capability of the
normally-provided control software for the MR apparatus. For this
purpose, real-time capability means that data measured with the
sequence (in this case, the sequence comprises imaging and
navigator sequences) can be evaluated during the sequencing, and
the further course of the sequencing can be influenced by the
results of this evaluation, wherein the time period between
recording the data and influencing the further course is short in
comparison with the typical time constants of the respiratory
movement (in this case, particularly, the respiratory cycle of a
human being, which amounts to between 3 and 10 seconds).
[0015] The main problem with the acceptance/rejection algorithm is
that the respiration of the patient frequently varies during the
examination. The variations in the respiratory movement can be,
thereby, such that the respiratory positions within the once
specified acceptance window are rarely, or no longer, detected.
This leads to extended acquisition periods and can even lead to the
measurement not being completed at all in the normal manner.
[0016] The most important algorithm, by far, that addresses this
problem is "Phase Ordering With Automatic Window Selection" (PAWS),
which is described, for example, in the article by P. Jhooti, P. D.
Gatehouse, J. Keegan, N. H. Bunce, A. M. Taylor, and D. N. Firmin,
"Phase Ordering With Automatic Window Selection (PAWS): A Novel
Motion-Resistant Technique for 3D Coronary Imaging," Magnetic
Resonance in Medicine 43, Pages 470-480 (2000) and in the US
patent, U.S. Pat. No. 7,039,451 B1. PAWS finds a final acceptance
window during the runtime, and can thus react in a flexible manner
to a changing respiration. A further goal of PAWS is to ensure a
certain degree of "phase-encode ordering" (or in short, "phase
ordering"). This means that adjacent lines in the k-space are
acquired in similar respiration states. In particular, a variation
in the respiratory state during acquisitions in the vicinity of the
k-space center, which is particularly sensitive to movement, is to
be avoided. PAWS was developed for a 3D Cartesian acquisition
technique. The ky-kz array system used for this acquires a complete
kx-kz plane of the 3-dimensional k-space following each navigator.
The modulation of the k-space signal along the kz axis resulting
from the transcendental state after interrupting the stationary
steady state by the navigator (as well as potential activated
preparation pulses, or the waiting for a further physiological
signal, such as an EKG trigger) on the kx-kz plane, is therefore
smooth. Discontinuations may arise in the ky axis as a result of
residual movement, which can be manifested in the image as
artifacts and blurring along the first phase encoding axis ky. This
does not only apply when the segment border exists in the vicinity
of the k-space center. Peristaltic movements, as well, which are
not detected by the respiratory sensor, can lead to artifacts in
the images.
[0017] PAWS exists in different variants, known as "bin" variants.
In PAWS, the width of the final acceptance window is established.
In contrast to the acceptance/rejection algorithm, the breathing
positions that this acceptance window includes are automatically
found at run time. The k-space filling takes place in clusters. A
cluster (in the original work the term "bin" was used instead of
cluster) is characterized by a breathing position range (an
acceptance range) and includes all k-space lines that have already
been measured after a breathing position has been measured in the
breathing position range associated with the cluster. In the n-bin
variant of PAWS, a breathing position range whose width is equal to
the acceptance window is covered by n successive clusters.
[0018] Furthermore, a starting position in the k-space is assigned
to each cluster, wherein the number of different starting positions
is n. Different starting positions are assigned to clusters with
adjacent respiratory positions where n>1. As soon as a
respiratory position assigned to a cluster is measured with the
navigator, the measurement of a k-space line that has not yet been
measured within said cluster is initiated. The decision regarding
which k-space lines still to be measured are selected takes into
consideration, as a whole, the already acquired k-space lines of
adjacent clusters as well. By way of example, a still missing
k-space line is selected such that an arbitrary group of n adjacent
clusters is complete to the greatest degree possible, wherein the
arbitrary group of n adjacent clusters contains the cluster to
which the current measured respiratory position is assigned; i.e.
the group of n adjacent clusters comprising the largest possible
number of different k-space lines. As soon as an arbitrary group of
n adjacent clusters comprises all of the k-space lines that are to
be measured, the measurement is stopped, because the overall
variation in the respiratory position is limited in these
measurement data, thereby, to the acceptance window.
[0019] The n different starting points and clusters of the n-bin
variation of PAWS normally result in n segments in the k-space. For
this, each segment consists of adjacent k-space lines. The
variations to the respiratory positions within a segment measured
with the navigator correspond to the position range assigned to a
cluster (in the original work, the term "bin size" is used), and
thus one n.sup.th of the acquisition window. The variation to the
respiratory position is greater over the course of the entire
k-space, and has an upper limit as a result of the specified
acceptance window. The lines belonging to the same segment are
measured during similar respiratory states. Thus, the modulation of
the signal changes with the respiration at the segment borders. As
a result, position jumps occur at the segment borders. An aim of
the different bin-variations of PAWS is to displace the segment
borders away from the movement sensitive k-space center. Another
aim is to obtain a greater degree of efficiency.
[0020] In the previously mentioned article by Jhooti et al., as
well as in the follow-up work by P. Jhooti, P. Gatehouse J. Keegan,
A. Stemmer, D. Firmin: "Phase ordering with Automatic Window
Selection (PAWS) with Half Fourier for Increased Scan Efficiency
and Image Quality;" Proc. Intl. Soc. Mag. Reson. Med. 11 (2004);
Page 2146, the 1-bin, 2-bin, 3-bin, and 4-bin variations are
compared with one another. The result of this comparison shows that
the 1-bin and the 2-bin variations of PAWS are the most efficient,
i.e. for a given width of the acceptance window, the measurements
are completed most quickly. The 1-bin variation is discarded
because it does not allow for "phase ordering," the 4-bin variation
(and higher) is discarded due to lower efficiency. The 3-bin
variation is less efficient than the 2-bin variation. The reason
for this is the unidirectional growth direction of the cluster with
starting positions at the left and right k-space edges. As soon as
the gap between one of these peripheral clusters and the central
cluster (with a starting position in the k-space center, and a
bidirectional growth direction) is closed, then said clusters
continue to grow until the gap between the other peripheral
clusters and the central cluster is closed, as soon as a
respiratory position is measured that is assigned to the first
peripheral cluster. This normally leads to multiple k-space lines
acquired at the cluster borders (segment borders). This problem
does not exist with the 2-bin variation. In this variation, every
second cluster grows in a unidirectional manner from the left-hand
k-space edge, through the k-space center, toward the right-hand
k-space edge, and the remaining clusters grow in a unidirectional
manner from the right-hand k-space edge, through the k-space
center, toward the left-hand k-space edge. The measurement is
complete as soon as two adjacent clusters (with opposite growth
directions) "meet." However, with a symmetrical scanning of the
k-space, as is the case with the 2-bin variation, the cluster
border frequently lies in the vicinity of the k-space center, which
is particularly sensitive to movement, which may lead to strong
image artifacts. The probability of cluster borders lying in the
vicinity of the k-space center is substantially lower with the use
of partial Fourier (i.e. an asymmetric scanning of the
k-space).
[0021] Of practical relevance, therefore, are the so-called 2-bin
and 3-bin versions of PAWS, wherein, with symmetrical scanning, the
3-bin variation is preferred, and with asymmetric scanning, the
2-bin variation is preferred. This analysis is based on a 2-bin
variation, in which the starting position alternates between the
left-hand and right-hand k-space edges of adjacent clusters.
Accordingly, the clusters grow, respectively, from the starting
positions assigned thereto, firstly toward the k-space center.
[0022] It is noted again that only a single breathing position is
associated with a cluster in some jobs. The width of the final
acceptance window then amounts to n-times the resolution of the
breathing signal. In this alternative formulation, a more flexible
selection of the acceptance window is achieved in that the
breathing position measured with the sensor is initially coarsened,
such that n-adjacent resulting breathing positions cover a
breathing range that corresponds to the width of the acceptance
window.
[0023] Three modifications of the 3-bin PAWS algorithm are known
from Nuval et al., "An improved real-time navigator gating
algorithm for reducing motion effects in coronary magnetic
resonance angiography"; Journal of X-Ray Science and Technology 11
(2003), P. 115-123 and A. Nuval et al., "Refined PAWS Algorithms
for 3D Coronary MRA". Proc. Intl. Soc. Mag. Reson. Med. 11 (2003),
P. 1625:
[0024] a) In the original 3-bin PAWS variant, clusters with start
position at the left k-space edge, in the k-space center and at the
right k-space edge alternate cyclically. In the modified version,
the start position alternates cyclically between left k-space edge,
in the k-space center, right k-space edge and k-space center again.
A start position in the k-space center is accordingly assigned to
every second cluster. Position jumps at the cluster boundaries that
are twice as large as the acceptance range assigned to a cluster
are avoided with this modification. However, this modification also
reduces the number of cluster combinations in which k-space can be
completed. The efficiency is thus reduced.
[0025] b) The termination criterion is tightened such that the
central cluster must have acquired at least 30% of k-space
symmetrically around the k-space center. The goal of this
modification is to avoid cluster boundaries near the k-space
center. This modification also extends the measurement time in
general, and therefore reduces the efficiency.
[0026] c) A histogram of the occurring breathing positions is
created with the aid of a prescan. The breathing position occurring
most frequently during the prescan is assigned to a central
cluster. This modification also reduces the probability of a
segment boundary near the k-space center. However, the efficiency
is reduced further by the prescan that is now necessary. Moreover,
the information obtained with the aid of a prescan can only be
transferred to the actual scan in the case of a regular
respiration. The integration of prescan information with the actual
PAWS therefore runs contrary to the idea of being robust with
regard to changing breathing patterns.
[0027] PAWS was originally developed for a ky-kz ordering scheme in
which all k-space lines are respectively acquired with a defined
value of the second phase coding gradient (in the direction of kz)
after acquisition of the breathing signal. The "phase ordering" is
accordingly also limited to a Cartesian k-space direction, which
can lead to intensified remaining movement artifacts in this
direction.
[0028] In a recent article, PAWS is combined with a known Radial
Phase Encoding (RPE) scheme (Christoph Kolbitsch, Claudia Prieto,
Jouke Smink and Tobias Schaeffter: "Highly Efficient Whole-Heart
Imaging Using Radial Phase Encoding-Phase Ordering With Automatic
Window Selection"; Magnetic Resonance in Medicine 66 (2011); P.
1008-1018). The respective data acquired after a navigator thereby
respectively have the same movement sensitivity. A special 2-bin
scheme is implemented. In the one bin set, radial spokes in k-space
are acquired in the clockwise direction; in the other bin set, they
are acquired in the counter-clockwise direction. The goal of this
scheme is to be able to repeatedly reconstruct the region of
interest (ROI) in different breathing phases.
SUMMARY OF THE INVENTION
[0029] An object of the invention is to provide a magnetic
resonance system and an electronically readable data storage medium
to implement a method in accordance with the invention with which
remaining movement sensitivities of the known PAWS methods are
reduced.
[0030] The invention is based on the following considerations.
[0031] Gating techniques are particularly important in connection
with Cartesian 3D gradient echo sequences. After each excitation
pulse, these sequences normally acquire a defined k-space line that
is specified by a ky value in the first phase coding direction and
a kz value in the second phase coding direction. This phase coding
line is read out once or multiple times at different echo times
(for example given use of a Dixon technique). The duration of such
an individual excitation (including signal coding and signal
acquisition) amounts to only a few milliseconds. Multiple phase
coding lines are therefore normally acquired after a single
navigator sequence to measure a physiological signal (such as the
breathing movement) and/or after the execution of a pre-switching
module to suppress unwanted signal contributions (for example to
suppress fat signals).
[0032] The set of all phase coding lines that are acquired after a
specific navigator sequence is called a "shot" in the following.
Since the magnetization is located in a transcendent state after
the interruption of the stationary equilibrium by the navigator
sequence (and possibly additional pre-switching modules, for
example for fat saturation), the chronological order of the ky-kz
lines within a shot determines the modulation of k-space, and
therefore the image quality.
[0033] Furthermore, it is known that the central k-space lines are
most movement-sensitive, and that the movement sensitivity of a
specific k-space line decreases with their distance from the
k-space center. Movement sensitivity is the tendency of the
formation of artifacts due to a movement in the examined
examination subject.
[0034] PAWS was originally developed for a ky-kz ordering scheme in
which all k-space lines within a shot were acquired with a specific
value of the phase coding gradient of the first phase coding
direction ky. In this ordering scheme, the number of excitations
per shot is equal to the number of phase coding steps Nz in the
second phase coding direction. The number of shots that are finally
accepted for image reconstruction is equal to the number of phase
coding steps Ny in the first phase coding direction. Accordingly,
the modulation of k-space as a result of the transcendent state
after interruption of the stationary equilibrium proceeds
exclusively along the second phase coding direction. In contrast to
this, remaining movement artifacts manifest along the first phase
coding direction. Furthermore, the movement sensitivity of a single
shot is characterized by its value of the phase coding direction
ky.
[0035] In order to reduce the movement sensitivity, and furthermore
to achieve a general description of PAWS, the following is
assumed:
[0036] a) The number of k-space lines (or more generally "views")
per shot is constant. The term "view" also includes non-Cartesian
k-space trajectories. For example, a view through the azimuthal
angle of a radial spoke and a kz coordinate in a phase coding
direction orthogonal to the radial plane can be described in a
radial 3D k-space trajectory or a spiral-shaped k-space trajectory
(optionally also with Cartesian sampling in a direction orthogonal
to the plane scanned in a spiral-shape).
[0037] b) The number of shots that is required for a complete
acquisition of k-space to be scanned is Ns.
[0038] c) A scalar that describes the neighborhood in k-space can
be assigned to each shot. A shot index ns in [0, . . . , Ns-1] is
ordered corresponding to this scalar.
[0039] d) There is a marked shot with shot index ns0 in [0, . . . ,
Ns-1] with maximum movement sensitivity. The movement sensitivity
accordingly increases in a range of [0, . . . , ns0] and decreases
in a range of [ns0, Ns-1].
[0040] In the ky-kz ordering scheme of the original work (Jhooti et
al.) that is described above, the number of shots Ns is equal to
Ny, and ky suggests itself as a scalar that describes the
neighborhood. Given a symmetrical acquisition of k-space, ky thus
suggests values in a range between -Ny/2 and Ny/2-1; the shot index
ns is obtained via the following conversion:
ns=ky+Ny/2.
The shot index ns0=Ny/2 with maximum sensitivity is situated
approximately in the middle of the value range.
[0041] The general description allows PAWS to be applied to a more
flexible ky-kz ordering scheme. What is understood by this is a
Cartesian k-space trajectory in which the individual k-space lines
are not acquired along one of the two Cartesian axes but rather
more or less along a radial line, whereby the movement sensitivity
is also reduced as already stated above.
[0042] In the method according to the invention for the acquisition
of a measurement data set of a breathing examination subject by
magnetic resonance technology, the measurement data set is acquired
with multiple shots that each includes a number Nv of k-space
trajectories (known as views). The acquisition of the measurement
data set in k-space includes the following steps.
[0043] A number Nv of views per shot is selected.
[0044] The number Ns of shots in order to completely fill k-space
to be scanned is then determined.
[0045] The views of the Ns shots are associated with Nv sectors in
k-space, such that approximately the same number of views are
arranged in each sector, and such that all views in a sector
respectively have a similar distance from the k-space center.
[0046] A respective view of each sector is associated with a
respective one of the Ns shots, corresponding to their orientation
in the kz-ky plane.
[0047] The views of each shot are scanned such that views that are
associated with the same sector and different shots respectively
assume the same (time) position within the shot.
[0048] The scanning of k-space according to the invention is robust
with regard to movements (even peristaltic movements, for example)
of the examination subject since movement along both Cartesian
directions is "blurred" in that the views of a shot are acquired
corresponding to their orientation in the kz-ky plane, and not (as
in the past) along a ky line, and therefore along only one phase
coding direction. The method is therefore less susceptible to
ghosting artifacts which arise as a result of remaining movement,
since (as already stated) scanning is blurred azimuthally.
[0049] The orientation can be determined in a simple manner in the
kz-ky plane via the azimuthal angle of a view in a polar coordinate
system. The association of the views of a sector with one of the Ns
shots can therefore take place corresponding to their azimuthal
angle in a polar coordinate system in the kz-ky plane.
[0050] The views associated with a shot are scanned in the same
order for each shot. This means that views that are associated with
a common sector are acquired in their shot at the same time after
the start of the respective shot. A smooth modulation of k-space
per shot therefore results, whereby additional ghosting artifacts
are avoided.
[0051] For example, the order for a scan of the views in each shot
can be chosen corresponding to the sectors with which the views of
the shot are associated. Since the views in one sector all have a
similar distance from the k-space center, for example, this
distance--with its direction relative to the k-space center--can be
used as an ordering criterion for the order, whereby the order with
which the views of each shot are scanned corresponds to a
neighborhood of the sectors.
[0052] A magnetic resonance system according to the invention has a
basic field magnet; a gradient field system; a radio-frequency
antenna; and a control device to control the gradient field system
and the radio-frequency antenna; and an image computer to receive
measurement signals acquired by the radio-frequency antenna, to
evaluate the measurement signals, and to create magnetic resonance
images. The control unit and the image computer are configured to
implement the method described above.
[0053] A non-transitory, electronically readable data storage
medium according to the invention has electronically readable
control information stored thereon, this control information
causing the inventive method to be executed when the data medium is
loaded in a control device of a magnetic resonance system.
[0054] The advantages and embodiments indicated with regard to the
method analogously apply to the magnetic resonance system and the
electronically readable data medium.
BRIEF DESCRIPTION OF THE DRAWINGS
[0055] FIG. 1 schematically illustrates a magnetic resonance system
according to the invention.
[0056] FIG. 2 shows an example of an association according to the
invention of views with sectors and shots, as it can be used in
connection with the PAWS algorithm that is explained in generalized
form with regard to FIG. 5.
[0057] FIG. 3 is a flowchart for an ordering according to the
invention for entering data into (scanning) k-space in sectors and
shots, as can be used in connection with the PAWS algorithm that is
explained in generalized form with regard to FIG. 5.
[0058] FIG. 4 is a flowchart of a generalized PAWS method.
[0059] FIG. 5 is an exemplary comparison of the PAWS ordering
scheme of the original work using the ky coordinate with a
generalization according to the shot index for a 3-bin PAWS
algorithm.
[0060] FIG. 6 is an exemplary comparison of the previous 2-bin PAWS
with an optimized 2-bin PAWS that uses the generalized PAWS method
described herein.
[0061] FIG. 7 is a flowchart of the optimized 2-bin PAWS.
DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0062] FIG. 1 schematically illustrates a magnetic resonance
apparatus 5 (a magnetic resonance imaging or tomography device). A
basic field magnet 1 generates, a temporally constant strong
magnetic field for the polarization or alignment of the nuclear
spin in a region of an examination subject U, such as a portion of
a human body that is to be examined, lying on a table 23 in order
to be moved into the magnetic resonance apparatus 5. The high
degree of homogeneity in the basic magnetic field necessary for the
magnetic resonance measurement (data acquisition) is defined in a
typically sphere-shaped measurement volume M, in which the portion
of the human body that is to be examined is placed. In order to
support the homogeneity requirements temporally constant effects
are eliminated by shim-plates made of ferromagnetic materials are
placed at appropriate positions. Temporally variable effects are
eliminated by shim-coils 2 and an appropriate control unit 27 for
the shim-coils 2.
[0063] A cylindrically shaped gradient coil system 3 is
incorporated in the basic field magnets 1, composed of three
windings. Each winding is supplied by a corresponding amplifier
24-26 with power for generating a linear gradient field in a
respective axis of a Cartesian coordinate system. The first partial
winding of the gradient field system 3 generates a gradient G.sub.x
in the x-axis, the second partial winding generates a gradient
G.sub.y in the y-axis, and the third partial winding generates a
gradient G.sub.z in the z-axis. Each amplifier 24-26 has a
digital-analog converter (DAC), controlled by a sequencer 18 for
the accurately-times generation of gradient pulses.
[0064] A radio-frequency antenna 4 is located within the gradient
field system 3, which converts the radio-frequency pulses provided
by a radio-frequency power amplifier into a magnetic alternating
field for the excitation of the nucleii by tipping ("flipping") the
spins in the subject or the region thereof to be examined, from the
alignment produced by the basic magnetic field. The radio-frequency
antenna 4 is composed of one or more RF transmitting coils and one
or more HF receiving coils in the form of an annular, linear or
matrix type configuration of coils. The alternating field based on
the precessing nuclear spin, i.e. the nuclear spin echo signal
normally produced from a pulse sequence composed of one or more
radio-frequency pulses and one or more gradient pulses, is also
converted by the RF receiving coils of the radio-frequency antenna
4 into a voltage (measurement signal), which is transmitted to a
radio-frequency system 22 via an amplifier 7 of a radio-frequency
receiver channel 8, 8'. The radio-frequency system 22 furthermore
has a transmitting channel 9, in which the radio-frequency pulses
for the excitation of the magnetic nuclear resonance are generated.
For this purpose, the respective radio-frequency pulses are
digitally depicted in the sequencer 18 as a series of complex
numbers, based on a given pulse sequence provided by the system
computer 20. This number series is sent via an input 12, in each
case, as real and imaginary number components to a digital-analog
converter (DAC) in the radio-frequency system 22 and from there to
the transmitting channel 9. The pulse sequences are modulated in
the transmitting channel 9 to a radio-frequency carrier signal, the
base frequency of which corresponds to the resonance frequency of
the nuclear spin in the measurement volume. The modulated pulse
sequences of the HF transmitter coil are transmitted to the
radio-frequency antenna 4 via an amplifier 28.
[0065] Switching from transmitting to receiving operation occurs
via a transmission-receiving switch 6. The RF transmitting coil of
the radio-frequency antenna 4 radiates the radio-frequency pulse
for the excitation of the nuclear spin in the measurement volume M
and scans the resulting echo signals via the HF receiving coils.
The corresponding magnetic resonance signals obtained thereby are
demodulated to an intermediate frequency in a phase sensitive
manner in a first demodulator 8' of the receiving channel of the
radio-frequency system 22, and digitalized in an analog-digital
converter (ADC). This signal is then demodulated to the base
frequency. The demodulation to the base frequency and the
separation into real and imaginary parts occurs after digitization
in the spatial domain in a second demodulator 8, which emits the
demodulated data via outputs 11 to an image processor 17. In an
image processor 17, an MR image is reconstructed from the
measurement data obtained in this manner through the use of the
method according to the invention, that includes computation of at
least one disturbance matrix and the inversion thereof, in the
image processor 17. The management of the measurement data, the
image data, and the control program occurs via the system computer
20. The sequencer 18 controls the generation of the desired pulse
sequences and the corresponding scanning of k-space with control
programs, in particular, in accordance with the method according to
the invention. The sequencer 18 controls accurately-timed switching
(activation) of the gradients, the transmission of the
radio-frequency pulse with a defined phase amplitude, and the
reception of the magnetic resonance signals. The time base for the
radio-frequency system 22 and the sequencer 18 is provided by a
synthesizer 19. The selection of appropriate control programs for
the generation of an MR image, which are stored, for example, on a
DVD 21, as well as other user inputs such as a desired number n of
adjacent clusters, which are to collectively cover the desired
k-space, and the display of the generated MR images, occurs via a
terminal 13, which includes units for enabling input entries, such
as, e.g. a keyboard 15, and/or a mouse 16, and a unit for enabling
a display, such as, e.g. a display screen.
[0066] FIG. 2 shows an example of a new k-space array system. A
kz-ky plane is shown, in which views (depicted by circles filled
with various patterns) are disposed in a Cartesian manner.
[0067] FIG. 3 shows a flow chart for arranging the k-space that is
to be scanned in sectors and shots.
[0068] For this purpose, first a number Nv of views, which are to
be acquired for each shot, is selected (block 31). This occurs, for
example, by means of an input by a user at a terminal 13 of a
magnetic resonance apparatus 5. The selection of the views per shot
can be freely selected with, for example, the use of a navigator
for determining the respiratory signal as a physiological signal,
as well as the temporal resolution of the respiratory signal,
because after a navigator, a shot with the selected number of views
is acquired. From the total number of the views to be measured,
(which is determined by means of, among other factors, the
resolution, which has, in turn, been selected by the user) and the
number Nv of views per shot is therefore established by the number
Ns of shots which are required in order to fully scan the k-space
that is to be scanned (block 303).
[0069] The views in the k-space are subdivided into sectors S1, S2,
S3, S4, S5, S6, S7, S8 (block 305). Views which are assigned to the
same sector are each filled with the same pattern. In addition, the
borders of the sectors S1, S2, S3, S4, S5, S6, S7, S8 are indicated
by thin broken lines.
[0070] The number of different sectors is the same as the number of
views per shot, and, e.g. a user-defined parameter. In the depicted
example, the number of sectors and the views per shot equals
eight.
[0071] The number of views per sector is the same as the number of
shots Ns. In the example, the number of views in each sector, and
therefore the number of shots, equals 49. Views which are assigned
to the same sector have a similar spacing from the k-space center,
and are located in the same hemisphere (in the example in FIG. 2,
the first hemisphere is defined by ky>0, or (ky=0 and
kz.ltoreq.0)). Advantages derived therefrom are obtained with an
asymmetric recording of the k-space (partial Fourier).
[0072] Each shot acquires, thus, one view per sector. For this,
views of a specific sector are acquired at the same point in time
after the navigator sequence, or after the starting of the shots,
respectively.
[0073] All views are assigned to a shot, wherein in each sector a
view is assigned to a specific shot (block 307). The assignment of
the individual views of a sector to a specific shot occurs in
accordance with their orientation in the kz-ky plane, for example,
in accordance with their azimuth angle in a polar coordinate
system. This array results in a smooth modulation of the k-space
(resulting from the transcendental state after an interruption of
the stationary steady state) along the quasi-radial scanning
direction.
[0074] As an example, three shots are depicted in FIG. 2 by means
of thick series of arrows. With the acquisition of measurement
data, the views assigned to the same sector and to different shots
each assume the same position within the shots (block 309). As one
sees in the example in FIG. 2, the sequence for the acquired views
for each shot corresponds respectively to the sectors thereof,
which, in the depicted case, are from S1 to S2 to S3 to S4 to S5 to
S6 to S7 to S8. The shots thus proceed in quasi-radial k-space
trajectories, in this case from the right edge to the left edge of
the k-space that is to be scanned.
[0075] The array system has the advantage, in comparison with the
array system previously used in conjunction with the PAWS
technique, that it is less susceptible to ghost artifacts resulting
from residual movement, because these are smeared at the azimuth.
Furthermore, the array allows for a free selection of the views per
shot, and therefore the temporal resolution of the physiological
respiratory signal recorded with the navigator sequence.
Furthermore, it is compatible with so-called elliptical scanning,
in which the views in the peripheral corners of the k-space,
particularly in the ky-kz plane, having a relatively low
information content, are not acquired, in favor of a shorter
measurement time period, as is also the case in the depicted
example. Furthermore, the intended distribution of the k-space
sectors is compatible with a varying density of the k-space
scanning, as is the case, for example, with parallel imaging with
auto-calibration.
[0076] The generalized PAWS description enables the array system
described above to be implemented together with PAWS, in that the
shots, as already mentioned above, are assigned a shot index for
the views distributed in the sectors, which fulfills the
suppositions c) and d) given above.
[0077] For this, the following approach can be used:
[0078] First, each shot exhibits an azimuth angle .phi. e.g.
between [-.pi., .pi.]. For this, one can, e.g., use the average
azimuth angle of the views per shot (arctan2 (ky, kz)) in the first
hemisphere, or the azimuth angle of the views in one of the
sectors, advantageously one of the sectors lying in the vicinity of
the k-space center, such as the sector S4 in FIG. 2. This azimuth
angle is a suitable scalar, and therefore array criteria, for the
shots, which describes a neighborhood of the shots. It is important
that one realizes that it is not necessary for the allocation shot
.fwdarw. scalar to be reversible. The shot index ns0 is set
advantageously to be equal to the shot index of the shots acquiring
the k-space center. In the example in FIG. 2, this shot has an
azimuth angle .phi.=0, and thus lies in the center of the
evaluation range of ns [0, Ns-1].
[0079] In a truly radial trajectory, all shots have the same degree
of sensitivity to movement. Nevertheless, one can still use the
generalized PAWS description. The selection of the ns0 is then
free.
[0080] With the determination by means of the above assumptions a)
through d), the PAWS algorithm can thus be applied to arbitrary
2-dimensional Cartesian ky-kz ordering scheme and k-space
trajectories in that the ky coordinate of the original work (Jhooti
et al.) is replaced by the shot index ns. For example, this occurs
simply by the shot index ns=0 being assigned as a start position to
a cluster ("bin" in the original work) with a start position on the
left k-space side (kymin in the original work), and the shot index
ns=Ns-1 being assigned as a start position to a cluster with start
position on the right k-space edge (kymax in the original work),
and the shot index ns=ns0 with maximum movement sensitivity being
assigned as a start position to a cluster with start position in
the k-space center (in the original work). An example of such a
conversion from the ordering scheme of the original work using the
ky coordinate (above) to an ordering scheme according to the shot
index for a 3-bin PAWS algorithm (below) is shown in FIG. 5. The
breathing position AP is plotted on the vertical.
[0081] A workflow diagram of a generalized PAWS method is
schematically shown in FIG. 4.
[0082] A flow chart for a generalized PAWS method is depicted in
FIG. 4.
[0083] For this, first, as is normal with PAWS, the number n is
selected, which indicates the number of adjacent clusters that
should, collectively, completely fill the k-space, in order to
obtain a complete measurement data set, which does not exceed a
given overall variation to the respiratory position during the
measurements (block 401).
[0084] Each of the Ns shots is assigned a shot index ns.epsilon.[0;
. . . ; ns0; . . . ; Ns-1] (block 403), as described above, wherein
the assignment occurs such that the shot indices ns are arranged
such that the sensitivity of the shots with respect to movement by
the examination subject increases from shot index ns=0 up to shot
index ns=ns0, and in turn, decreases from shot index ns=ns0 to shot
index ns=Ns-1.
[0085] The measurement is initiated with a navigator measurement
for determining a respiratory signal and therefore a momentary
respiratory position (block 405).
[0086] For this purpose, the shot that is to be measured after the
navigator measurement is assigned to a cluster, in the PAWS method,
in a typical manner, corresponding to the respiratory signal
measured with the navigator measurement (block 407).
[0087] After it has been determined by the navigator measurement,
to which cluster the following shot that is to be measured is
assigned, a shot index is determined for the shot to be measured,
in relation to the already acquired shot indices for the previously
selected cluster and its adjacent clusters (block 409). If no shot
has yet been acquired in the cluster to which the measured shot is
to be assigned, the shot having the shot index corresponding to the
starting position assigned to said cluster is selected (block
409).
[0088] The corresponding shot with the selected shot index is
acquired as the shot that is to be measured (block 411).
[0089] If, after the last acquisition of a shot, in a given number
n of adjacent clusters, all of the shots with all of the Ns shot
indices have been acquired (query 413) then the measurement is
complete (block 415).
[0090] Accordingly, with the generalized PAWS method as well, a
so-called peripheral cluster having a starting position 0 grows
toward the k-space center, in that it selects the next, not yet
acquired, larger shot index, and a peripheral cluster having a
starting position Ns-1 grows in that it selects the next smaller,
not yet acquired shot index.
[0091] The central cluster (having the starting position ns0)
selects from the n possible cluster combinations the cluster that
is complete to the greatest extent, i.e. the cluster already
comprising the most shots having different shot indices, and then
grows toward a smaller, or larger, respectively, shot index,
depending on whether the quantity formed by shots having ns
.ltoreq.ns0, or, respectively, the quantity formed by shots having
ns.gtoreq.ns0, which are not yet acquired from the cluster
combination, has more elements. As soon as an arbitrary group of n
adjacent clusters comprises all of the shot indices [0, . . . ,
Ns-1] that are to be measured, the measurement is complete (block
415), because the overall variation in the respiratory position is
thus limited to the acceptance window. If there are still cluster
combinations of n adjacent clusters shots in which not all Ns shot
indices are comprised, then at block 405, the method is continued,
and a new navigator measurement is acquired, in each case, with a
new subsequent shot.
[0092] In the example in FIG. 2, adjacent views in the k-space each
belong to shots with a similar azimuth angle, and thus to shots
with adjacent shot indices. By this means, the generalized PAWS
algorithm ensures that adjacent views are acquired during similar
respiratory states. In the example in FIG. 2, the shot that
acquires the k-space center has the azimuth angle .phi.=0, and thus
lies in the center of the evaluation range of the shot indices. One
can thus assume that the results contained in the original work
(with regard to efficiency and remaining susceptibility to
movement) can also be directly applied to the proposed, more
flexible, ky-kz array system. With the described method, the PAWS
concept can be used with any arbitrary ky-kz array system and with
any arbitrary non-Cartesian k-space trajectories.
[0093] With the use of a navigator for recording the movement, the
stationary steady state of the magnetization is interrupted by the
execution of the navigator sequence. With the array system
according to the original work from Jhooti et al., this
interruption occurs, in each case, after the acquisition of Nz TR
intervals, wherein Nz is the number of phase encoding steps in the
second phase encoding direction. The temporal resolution of the
respiratory signal is thus linked directly to the spatial
resolution of the imaging sequence along the second phase encoding
direction. However, the respiratory signal measured with the
navigator is only valid for a limited period of time, which is
short in relation to the respiratory interval. This means that the
array system used in the original work by Jhooti et al. inherently
limits the maximum resolution in one of the two Cartesian axes.
With the use of the proposed, generalized PAWS algorithm, having an
array system such as that described, in particular, in reference to
FIG. 2, a limitation of this type does not exist, because the
number of views per shot, and thus the temporal resolution of the
respiratory signal, can be freely selected. This advantage is
particularly important, because the goal of respiratory gated
measurements is frequently to avoid the inherent resolution
limitation to measurements made while holding one's breath,
resulting from the limited ability of the patient to hold its
breath for longer periods of time.
[0094] The problem of the limited temporal validity of the
navigator signal can be avoided in part through the use of a
1-dimensional centric array system along the kz axis. An array
system of this type starts in the k-space center, and acquires
alternating views with positive and negative values for kz, in such
a manner that the absolute moment of the phase encoding steps grows
in a continuous manner. This centric array system has, however, the
disadvantage that it can lead to artifacts resulting from
turbulences as a result of the larger phase encoding jumps between
the TR intervals.
[0095] As an alternative to the normal n-bin PAWS method, in the
following an optimized 2-bin PAWS method shall be presented.
[0096] FIG. 6 shows, by way of example, a comparison of the
previous 2-bin PAWS with a new, optimized 2-bin PAWS, wherein a
prior 2-bin PAWS is depicted at the top, and the new 2-bin PAWS is
depicted below.
[0097] As is described in the original work (Jhooti et al.), in the
original 2-bin PAWS variation the starting position of adjacent
clusters alternates between the left-hand and the right-hand
k-space edge. In the upper part of FIG. 6, clusters having an
even-numbered index are assigned the right-hand starting position,
and clusters with an odd-numbered index are assigned the left-hand
starting position. This corresponds in the generalized depiction,
described herein, to an alternation between ns=0 and ns=Ns-1.
Accordingly, a cluster with the starting position ns=0 grows, in
that it selects the smallest shot index that has not yet been
acquired from the cluster. A cluster with a starting position
ns=Ns-1 grows in that it selects the largest shot index that has
not yet been acquired from the cluster. In the following, a shot
index ns shall always be referred to, even if the original array
can be used in accordance with the ky coordinates. The width of the
respiration position range assigned to each cluster normally
corresponds to half of the acceptance window (AF). The measurement
is complete as soon as two arbitrary adjacent clusters have
collectively acquired all shots. This is the case in the example in
FIG. 6 for the clusters c4 and c5. In terms of imaging, a cluster
"growing from the left side" (starting position ns=0) and one of
the two adjacent clusters "growing from the right side" (starting
position ns=Ns-1) meet, such that both clusters, collectively, span
the overall value range [0, . . . , Ns-1]. To the extent that these
two clusters comprise nearly the same number of shots, frequently a
cluster border (marked with an oval drawn with a broken line in
FIG. 6.) is then obtained in this connection in the movement
sensitive region surrounding the k-space center.
[0098] In the optimal 2-bin PAWS implementation presented here, two
cluster types also alternate. The one cluster type has the shot
having the maximum movement sensitivity ns=ns0 as the starting
position and shall be referred to in the following as the central
cluster. The other cluster type does not have a clear starting
position, and shall be referred to in the following as a peripheral
cluster. In the example in FIG. 6, clusters with an odd-numbered
index are central clusters, and clusters with an even-numbered
index are peripheral clusters. The starting position of a
peripheral cluster is either ns=0, or ns=Ns-1, or, respectively,
either the right-hand or the left-hand k-space edge in the ky
coordinates, wherein the actual starting position is first decided
on during the runtime. A peripheral cluster grows, independently of
its starting position, either from the largest shot index not yet
belonging to the cluster, downward toward ns0, or ky=0,
respectively, or from the smallest shot index not yet belonging to
the cluster, upward toward ns0 or, ky=0, respectively. The decision
as to which direction of growth shall currently be preferred,
occurs in turn during the runtime. This is schematically depicted
in a flow chart in FIG. 7.
[0099] If, as has already been described above, a respiratory
position is measured by means of a navigator measurement (block
701, corresponding to block 405 in FIG. 4), which lies in the
respiratory position range of a peripheral cluster cn, in
accordance with the normal array used in PAWS (block 407 in FIG.
4), then it is next queried whether the cluster combination
cn-c(n-1) or cn-c(n+1) is closer to completion (blocks 705 and 707.
For this purpose, the shots already acquired in the clusters
cn-c(n-1) and cn-c(n+1) are first counted, and these are stored
with the respective number M- or M+ (block 705). The adjacent
cluster cx, which, together with the peripheral cluster cn, is
closest to completion, is selected according to these numbers M- or
M+, wherein the cluster c(n-1) is selected if M+ is greater than
M-. The clusters are labeled in the normal fashion, corresponding
to their respiratory position range (cluster cn corresponds to the
n.sup.th respiratory position). Accordingly, c(n-1) and c(n+1) are
central clusters, and the next two neighbors thereof are peripheral
clusters cn. In this manner, the cluster c(n-1) is selected if the
cluster combination cn-c(n-1) is closer to completion, and
otherwise, the cluster c(n+1) is selected.
[0100] Next, the number of shots Mlow, having an index in the range
[0, . . . , ns], which have not yet been acquired from the two
clusters (cn and the selected cx), and the number of shots Mhigh,
having an index in the range [ns, . . . , Ns-1], which have not yet
been acquired from the two clusters, are counted (block 709). If
the cluster cn, which is assigned to the last measured respiratory
position, as in the given case, is a peripheral cluster ("y" in
query 711), then the peripheral cluster cn grows from its smallest
not yet acquired shot index toward ns0, or ky=0, respectively, if
Mlow is greater than Mhigh ("y" in query 713), in which the
smallest, not yet acquired, shot index is acquired (block 715);
otherwise, it grows from its largest not yet acquired shot index
toward ns0, or ky=0, respectively ("n" in query 713), in which the
largest, not yet acquired shot index, is acquired (block 715).
[0101] A peripheral cluster spans, therefore, in general, two
contiguous index ranges. The one starts at the smallest shot index
ns=0 (or the left-hand k-space edge), and grows toward larger shot
indices. The other starts at the largest shot index ns=Ns-1 (or the
right-hand k-space edge), and grows toward smaller shot indices.
Alternatively, one can also refer to the index range for peripheral
clusters being continued at the range borders in a periodic or
cyclical manner.
[0102] The decision process runs in a similar manner, when the last
measured respiratory position lies in the respiratory position
range of a central cluster cn.
[0103] As explained above, it is checked to see which of the
cluster combinations, cn-c(n-1) and cn-c(n+1), is closer to
completion, and this cluster combination is selected (blocks 705
and 707). Next, the number of shots Mlow, having an index in the
range [0, . . . , ns], which are not yet acquired from the two
clusters, as well as the number of shots Mhigh, having an index in
the range [ns, . . . , Ns-1], which are not yet acquired from the
two clusters, are counted (block 709). If the cluster cn, which is
assigned to the last measured respiratory position, is a central,
as is assumed in the present case, ("n" in query 711), then the
central cluster cn, if Mlow is greater than Mhigh, grows from its
smallest already acquired shot index toward ns=0, or kmin,
respectively ("n" in query 719), in which the largest of the not
yet acquired shot indices is acquired, which is smaller than the
smallest already acquired shot index (block 723). Otherwise (Mhigh
is greater than Mlow) ("y" in query 719) the cluster cn grows from
its largest already acquired shot index toward Ns-1, or kmax,
respectively, in which the smallest of the shot indices that is
larger than the largest already acquired shot index is acquired
(block 721).
[0104] In any case, after an acquisition of a shot in one of the
blocks 715, 717, 721 and 723, it is checked in query 725 whether
all Ns desired shot indices are already acquired in the selected
cluster combination. If this is the case ("y" in query 725), the
measurement is complete (block 727), and can be stopped; if not
("n" in query 725), then the process is continued with a new
navigator measurement. In this manner, the termination criteria
remains unchanged with respect to the original version of PAWS; as
soon as an arbitrary group of two adjacent clusters (2-bin) has
acquired all of the shot indices that are to be measured, the
measurement is terminated, because the overall variation of the
respiratory position is limited thereby to the acceptance
window.
[0105] In the workflow diagram in FIG. 7 that summarizes the
algorithm just described, the typical symbols of set theory are
used:
[0106] {.} . . . designates a set
[0107] {xs| . . . } designates the set of all shot indices xs "for
which . . . is valid"
[0108] .epsilon. means "is an element of"
[0109] #{.} . . . designates the number of elements of the set
[0110] . . . logical symbol for "and"
[0111] v . . . logical symbol for "or"
[0112] This optimal 2-bin version of PAWS unites the high degree of
efficiency of the original 2-bin PAWS version with the reduced
artifact susceptibility of the original 3-bin variation. The new
algorithm actively pushes the segment borders away from
particularly movement sensitive k-space centers toward k-space
peripheries.
[0113] In the following comparison of the various PAWS varieties,
it is assumed that the overall width of the acceptance window is
given. With an n-bin variation, this acceptance window is spanned
by n successive clusters of the respiratory position range. As an
example, each cluster is assigned a respiratory position range, the
width of which corresponds to 1/n.sup.th of the acceptance window.
This differs in comparison, for example, with the Appendix A in the
MRM article by Jhooti et al., already cited above, in which the
width of the respiratory position range of a cluster is set to be
equal to the navigator resolution. With the latter approach, the
overall width of the acceptance window is n.times.the navigator
resolution, and increases with the number of bins. This makes it
difficult to carry out a fair comparison of different
bin-variations.
[0114] The efficiency of the new 2-bin variation, described above,
is optimal in the sense that as soon as a respiratory position, in
a range covered by two adjacent clusters, is measured Ns times, all
Ns shots are recorded, and thus the measurement can be stopped.
This property is shared by both the new 2-bin variation and the
original 2-bin variation, and is distinguished from the original
3-bin variation and the 3-bin variation from the writings by Nuval
et al., cited above.
[0115] In contrast to the original 2-bin variation, the probability
of cluster borders existing in the vicinity of the movement
sensitive k-space center is significantly reduced.
[0116] This is visible in FIG. 6, in which, as is normal, each line
corresponds to a cluster cn. These are disposed in the vertical
plane corresponding to their respiratory position range. In the
horizontal plane the phase encoding index is ky, or, respectively,
in the general depiction, applied as the shot index ns. The grey
shaded bars indicate the ky lines or shots, respectively, acquired
from a cluster. The upper portion of FIG. 11b refers to the MRM
articles by Jhooti et al. cited above, which represent the
selection of the phase encoding lines of the original 2-bin
variation at the end of the measurement. In the lower portion, the
corresponding presentation of the new 2-bin variation is depicted.
It can be seen that the number of times a specific respiratory
position is measured is the same in both plots. In the original
PAWS method depicted in the upper portion, the cluster border is in
the vicinity of the k-space center. In the new variation, depicted
in the lower portion, it is displaced to a significant degree
toward the periphery of the k-space. In FIG. 6, the cluster borders
are highlighted in each case with an oval drawn with a broken
line.
[0117] This problem is largest when the respiratory positions,
which are assigned to the two last clusters, are measured with
approximately the same frequency, and the central, particularly
movement sensitive, shot, having the shot index ns0, or ky=0,
respectively, lies precisely in the center of the index range. In
this case, the cluster border lies precisely in the k-space center
(ky=0). The new version deals with this particularly important case
in an optimal manner: the new cluster borders lie at ca. +25% and
+75% of the value range, and are thus maximally distanced from the
movement sensitive k-space center.
[0118] With an asymmetric scanning of the k-space as well, the new
2-bin variation presented herein functions in an optimal manner in
the sense that, with the given number of scans occurring for the
central cluster, the segment borders are distanced from the
central, particularly movement sensitive shot, having a shot index
ns0, or ky=0, respectively, to the maximum extent. Thus, the new
version, for all practical purposes, always functions better than
the original 2-bin version. The reason for this is that the
symmetrical distribution of the shots about the central shot ns0 is
actively incorporated in the decision process of the algorithm.
[0119] Then, and only then, if the number of scans which occur in
the final central cluster is less than the number of scans
occurring in the final peripheral cluster, a cluster border may
exist in the vicinity of the k-space center. In this case, this
border may lie closer to the k-space center than with the original
version. This case, however, is extremely unlikely with a
reasonable distribution of the acceptance window and a static
distribution of the respiratory position in the vicinity of the
most probable respiratory position, and has not been observed in
our numerous measurements made using the new 2-bin variation. By
means of an expansion, similar to the modification b), from one of
the to documents cited above by Nuval et al., this case can even be
entirely prevented: one restricts the termination criteria in such
a manner that the central final cluster must have either acquired a
minimum percentage of all shots Ns, or the peripheral final cluster
must have acquired all shots Ns. It can be seen that the
symmetrical distribution about the k-space center in the new 2-bin
algorithm is inherent thereto, and need not be stipulated (in
contrast to the 3-bin variations in the prior art).
[0120] It should also be mentioned that the borderline case of "no
respiration" is managed in an optimal manner with the new 2-bin
algorithm (as well as with the original version): all shots are
acquired from a single cluster, and thus there are no cluster
borders, regardless of whether this cluster is a central or a
peripheral cluster.
[0121] FIG. 8 shows, in a schematic manner, a flow chart for a
method for determining the respiratory phase.
[0122] For this purpose, a series of measured physiological signal
points (s.sub.1, s.sub.2, . . . , s.sub.n) is given, which
correspond to the respiratory positions. In the case of s.sub.n,
this can be, for example, the last respiratory position measured
with the navigator sequence at the point in time t.sub.n. s.sub.n-1
is the respiratory position measured directly prior to this at the
point in time t.sub.n-1, etc. The measurements are afflicted with
noise. Each measurement point s.sub.n can therefore be described
as:
s.sub.n=q.sub.n+n.sub.n,
[0123] wherein q.sub.n is the unknown, actual physiological state
(thus, in the example, the diaphragm position at the point in time
t.sub.n) and n.sub.n is likewise, the unknown noise.
[0124] Without limitation to the general condition, it is further
assumed that a local maximum of the series (q.sub.n) corresponds to
a state at the end of the exhalation (end-expiration state), and a
local minimum of the series (q.sub.n) corresponds to a state at the
end of the inhalation (end-inspiration state). This stipulation is
made in order to keep the description simple. If the signal course
is reversed, such as with the measurement of the chest size using a
respiratory belt, then all signal points s.sub.n can simply be
multiplied by -1, and one proceeds analogously.
[0125] The aim of the algorithm is to determine the respiratory
phase at an arbitrary point in time t.sub.n from the previous
measurements, without knowing the physiological signal points
(s.sub.n+1, s.sub.n+2, . . . ) at the later points in time n+1,
n+2, . . . . The respiratory phase is a triplet in this case, which
can assume the states {"unknown" (U), "exhalation" (E), and
"inhalation" (I)). For this, the state "unknown" is only assumed at
the beginning. As soon as it is abandoned, it is never again
obtained.
[0126] If the series (q.sub.n) were already known, this problem
would be trivial. From q.sub.n>q.sub.n-1 exhalation (expiration)
could be assumed, and q.sub.n<q.sub.n-1 would imply inhalation
(inspiration). In the case of q.sub.n=q.sub.n-1, the previous state
would be retained.
[0127] Another input in the algorithm is a threshold parameter
.DELTA.s, which-roughly speaking-distinguishes noise from a change
in the signal course resulting from the respiratory phase. .DELTA.s
can, for example, be determined by a one-time calibration
measurement of the standard deviation of the series (s.sub.n). In
our implementation, the empirical value .DELTA.s=4 mm is used, if
it is the case that the navigator measures a diaphragm
position.
[0128] The algorithm can be interpreted as a status machine (in
German, also referred to as a "finite automat," and in English as a
"finite state machine" (FSM)).
[0129] FIG. 8 illustrates the algorithm using a flow chart.
[0130] For this purpose, a physiological signal point s.sub.n (the
respiratory signal) is measured (block 801). If the measured signal
point s.sub.n is the first measured signal point (s.sub.n=s.sub.1),
then first, the state "unknown" is assigned ("y" in query 803).
Thus, the initial state of the respiratory phase is "unknown."
[0131] If the final state machine is in this state, then both a
prior maximum measured signal value s.sub.max, as well as a prior
minimum measured signal value s.sub.min as of the last change to
the respiratory phase, is recorded.
[0132] With the measurement of the first signal point s.sub.1, the
following initialization occurs (block 805):
Phase="unknown" (U); s.sub.min=s.sub.1; s.sub.max=s.sub.1.
[0133] Each new measurement s.sub.k where k>1 ("n" in query 803)
can be a transition in the respiratory phase, or trigger the state
"exhalation" (E) or "inhalation" (I). For this, first the momentary
respiratory phase, i.e. the respiratory phase to which the last
signal point was assigned, is queried (query 807).
[0134] According to the first measurement s.sub.1, as stated, the
respiratory phase "unknown" (U) is the momentary respiratory phase.
From here, a transition to the inhalation state (I) occurs, if
s.sub.k.ltoreq.S.sub.max-.DELTA.s ("y" in query 809). Thus, the
current respiratory phase is "inhalation" (I), and the minimum
measured signal value in this respiratory phase is updated to
s.sub.min=s.sub.k (block 811). s.sub.min is a variable in this
case, which indicates the minimal signal value as of the current
transition to the state "inhalation" (I).
[0135] If s.sub.k>S.sub.max-.DELTA.s ("n" in query 809), then a
transition to the state "exhalation" (E) occurs, if
s.sub.k.gtoreq.s.sub.min+.DELTA.s ("y" in query 813). Thus, the
current respiratory phase is "exhalation" (E) and the maximum
measured signal value is this respiratory phase is updated to
s.sub.max=s.sub.k (block 815). s.sub.max is a variable which
indicates the maximum signal value as of the current transition to
the state "exhalation" (E).
[0136] If neither the query 809, nor the query 813 is correct ("n"
in query 813), then s.sub.k does not trigger a state transition,
but instead the respiratory phase "unknown" is also the current
respiratory phase and the variables s.sub.min and s.sub.max are
updated:
s.sub.min=min(s.sub.min, s.sub.k); s.sub.max=max(s.sub.max,
s.sub.k).
[0137] If the finite state machine is in the state "exhalation" (E)
("E" in query 807), then with the measurement of the next signal
point s.sub.1, either the state "exhalation" (E) can be retained, a
transition to the state "inhalation" (I) occurs. For this, query
819 is carried out, which checks whether
[0138] s.sub.1.ltoreq.s.sub.max-.DELTA.s. If yes ("y" in query
819), then the respiratory phase currently changes to "inhalation"
(I), and the variable s.sub.min is updated to s.sub.min=s.sub.1
(block 821). If not ("n" in query 819), then the state "exhalation"
(E) is retained for the current respiratory phase, and the variable
s.sub.max is updated to s.sub.max=max(s.sub.max,s.sub.1) (block
823).
[0139] s.sub.min is a variable indicating the minimum signal value
as of the current transition to the state "inhalation" (I). The
initialization with the current measured signal point is
independent of whether the transition occurs from the state
"unknown" (U) or "exhalation" (E).
[0140] If the finite state machine is in the state "inhalation" (I)
("I" in query 807), then with the measurement of the next signal
point s.sub.m, either a transition to the state "exhalation" (E)
can occur, or the state "inhalation" (I) is retained. For this, the
query 825 is carried out, which checks whether
s.sub.m.gtoreq.s.sub.min+.DELTA.s. If yes ("y" in query 825), then
the respiratory phase currently changes to the state "exhalation"
(E), and the variable s.sub.max is updated to s.sub.max=s.sub.m
(block 827). If not ("n" in query 825), then the state "inhalation"
(I) is retained for the current respiratory phase, and the variable
s.sub.min is updated to s.sub.min=min(s.sub.min, s.sub.m) (block
829).
[0141] In this manner, each new signal point s.sub.n is assigned to
an unambiguous respiratory phase. This allocation is implicitly
dependent on the momentary state of the finite state machine and
the variables s.sub.min and/or s.sub.max from the previously
measured signal points, but not on the future s.sub.n+1, . . . ,
which are unknown in a prospective decision. As soon as the initial
state "unknown" has been abandoned once, then the prospective
gating can be initiated, taking into account the binary respiratory
phase.
[0142] It should be noted here, that a digital filtering or an
evening out of the series (sn) can be entirely eliminated here.
[0143] By means of this method, the respiratory phase in a
prospective gating method can be determined, also by means of
respiratory positions measured by means of navigator measurements,
and used in order to keep the actual variations to the respiratory
positions in the final measurement data that are used for the image
reconstruction to a minimum.
[0144] A flow chart of a method for the acquisition of a
measurement data set in individual measurements is depicted
schematically in FIG. 9, wherein, for each individual measurement,
a respiratory position and a respiratory phase are determined,
based on which it is decided whether the individual measurement is
to be recorded in a final measurement data set from which an image
data set is reconstructed.
[0145] For this, a physiological signal, the respiratory position,
is measured, and, for example, as was described above in reference
to FIG. 8, an associated respiratory phase is determined (block
901.1). Subsequently, a individual measurement for the measurement
data set that is to measured is acquired, which is assigned to the
previously determined respiratory phase and the measured
respiratory position (block 901.2).
[0146] If the assigned respiratory phase is "unknown" (U) ("y" in
query 903), the data of the individual measurement are discarded,
or not recorded in a final measurement data set, respectively, from
which an image data set for the examination subject that is to be
measured is reconstructed (block 905).
[0147] If the assigned respiratory phase is not "unknown" (U) ("n"
in query 903), then the measurement data are stored together with
their respiratory phase and respiratory position (block 907).
[0148] By means of the query 909, it is decided, based on the
respective assigned respiratory position and respiratory phase, and
by means of a selected algorithm, whether the acquired measurement
data are to be recorded in the final measurement data set ("y" in
query 909) and accordingly, stored in a possible final measurement
data set (block 913), or whether the acquired measurement data can
be discarded in block 911 ("n" in query 909). Depending on the
algorithm used, it is possible to store numerous possible final
measurement data sets, wherein the measurement is stopped when one
of these possible measurement data sets fully scans the desired
k-space.
[0149] If the measurement data stored up until this point in the
final measurement data set fully cover the entire k-space that is
to be acquired (query 915), then the process stops, and an image
data set is reconstructed from the final measurement data set
(block 917), and if not ("n" in query 915), then a respiratory
position and a respiratory phase are determined, and an associated
individual measurement is acquired. In this manner, respiratory
positions and respiratory phases are determined, and associated
individual measurements are acquired, until the k-space
corresponding to the examination subject that is to be examined has
been fully recorded.
[0150] Which criteria are to be used in query 909 depends on the
selected algorithm.
[0151] By way of example, an acceptance/rejection algorithm can be
selected in which the k-space that is to be recorded for the image
reconstruction is measured by segments in individual measurements
(block 901.2). Each measurement of a segment is assigned a
physiological data point (block 901.1). A physiological data point
is characterized thereby by means of a respiratory position and a
respiratory phase. A specific data segment is then accepted in
query 909, i.e. it is used later in block 917 for image
reconstruction, only if the respiratory position of the
physiological data point assigned thereto lies within an acceptance
window, and the respiratory phase of the physiological data point
assigned thereto assumes a given state, e.g. "inhalation." If this
is the case, then two final data sets can be stored thereby, which
each accept a different respiratory phase. In this case, the
measurement is stopped (block 917), if one of the two final
measurement data sets fully comprises the k-space that is to be
acquired.
[0152] The requirement of a concrete respiratory phase
distinguishes the presented method from the prior art. If one of
the two conditions is not met, the measurement of the segment is
repeated ("n" in query 915). The respiratory position of the
physiological data point is, for example, recorded, as is the case
in the prior art, with a navigator sequence, which is executed
either directly before or directly after the measurement of the
segment. In certain k-space trajectories (such as radial or spiral)
the respiratory position can also be extracted directly from the
measurement data. Such a method can be called a self navigated or
"self gated" method. The respiratory phase assigned to
physiological data point is determined, in contrast to this, from
the series of the previously occurring navigator measurements.
[0153] The type of k-space segmentation does not play a role in the
method. Because the navigator sequence frequently interrupts the
stationary steady state of the imaging sequence, one preferably
inserts the navigator at such positions in the sequencing in which
the stationary steady state is necessarily interrupted for other
reasons, e.g. due to a fat saturation pulse, or when waiting for
another physiological signal, such as an ECG trigger.
[0154] In another embodiment, a PAWS algorithm can be selected,
wherein the prerequisites for the imaging sequence, with which the
individual measurements are acquired, (as described above,
summarized once again) are:
[0155] a) An individual measurement is a shot, wherein a "shot" is
understood to mean all of the measurement data of the imaging
sequence, which are acquired following a specific navigator
sequence (and prior to the following navigator sequence). These
data are assigned to the respiratory position measured with the
navigator sequence, and the determined respiratory phase. In
general, one selects a constant time period for the length of the
shots, but this is not absolutely necessary.
[0156] b) The number of shots required for a complete recording of
the k-space that is to be scanned is Ns.
[0157] The prerequisites a and b are sufficient for automatically
establishing the final acceptance window. Should additional "phase
ordering" be carried out, the following two prerequisites c and d
are needed in addition:
[0158] c) Each shot can be assigned a scalar, which describes the
neighborhood in the k-space. The shot index ns in [0, . . . , Ns-1]
is classified in accordance with this scalar.
[0159] d) There is a distinctive shot having the shot index ns0 in
[0, . . . , Ns-1] having a maximum movement sensitivity.
Accordingly, the movement sensitivity increases in the range [0, .
. . , ns0], and decreases in the range [ns0, . . . , Ns-1].
[0160] For this, each navigator measurement is assigned a (scalar)
respiratory position and a binary respiratory phase ("inhalation,"
"exhalation"), or, respectively, a respiratory phase from the
triplet ("inhalation," "exhalation," "unknown"), as described
above. The prerequisite for this is that a navigator measurement,
which occurs temporally after the individual measurement, to which
the respiratory phase is to be assigned, is not necessary for
determining the respiratory phase. This can be obtained, for
example, with the method described in reference to FIG. 8.
[0161] One embodiment of the method, which uses a generalized PAWS
algorithm, presented above, is characterized by the following
properties:
[0162] a) A cluster is characterized by a respiratory position
range (or acceptance range) and a respiratory phase. It comprises
all shot indices (block 913), which have already been measured
after a respiratory position in the respiratory position range
assigned to the cluster, and the respiratory phase assigned to the
cluster have been measured.
[0163] b) There are two cluster sets. The first cluster set
consists of so-called inspiratory clusters (i.e. clusters assigned
the cluster phase "inhalation"). The second set consists of
so-called expiratory clusters (i.e. clusters having the respiratory
phase "exhalation" assigned thereto). Data with undetermined
respiratory phases are discarded.
[0164] c) The clusters are classified according to their
respiratory position range. The respiratory position ranges of
different clusters are disjunctive, and "gap-less." For each
measured respiratory position, there exists, therefore, exactly one
cluster in which this respiratory position falls within its
acceptance range. Furthermore, the respiratory position ranges of n
successive clusters each cover a respiratory position range, which
is coordinated to the desired final width of the acceptance
window.
[0165] d) As soon as an arbitrary group of n adjacent clusters
comprises a set of all of the shot indices [0, . . . , Ns-1] that
are to be measured (query 915), the measurement can be stopped
(block 917), because the overall variation to the respiratory
position is thus limited to the acceptance window, and all of the
measurement data belonging to the group of n adjacent clusters
during a respiratory phase have been measured.
[0166] The characteristics a)-d) are sufficient for an automatic
location of the final acceptance window. With the characteristic e
a phase ordering can additionally be obtained.
[0167] e) Each cluster is assigned a starting position in the form
of a shot index (also referred to as starting index ns_seed) in the
interval [ns0, . . . , Ns-1]. The cluster grows, starting from the
starting position, in such a manner that the index range covered by
the cluster is without gaps. Optimally, for this the index range is
continued at the range borders in a periodic or cyclical manner, as
described above in reference to the optimized 2-bin PAWS. The
decision of whether a cluster grows toward smaller or larger shot
indices takes into consideration, in general, the shot indices
already acquired from adjacent clusters. The aim is to select an
index in such a manner that an arbitrary group of n adjacent
clusters, which contains the cluster to which the currently
measured respiratory position is assigned, is complete to the
greatest extent possible, i.e. comprises the larges possible number
of different shot indices. The optimal distribution of the starting
positions between adjacent clusters depends on the number of
bins.
[0168] The embodiment described above requires that the final
accepted data be completely measured during the respiratory phase
"inhalation," be completely measured during the respiratory phase
"exhalation."
[0169] In another embodiment, which likewise uses a generalized
PAWS algorithm, presented above, an exception is allowed: In the
particularly quiet phase at the end of the exhalation, and with the
initiation of the inhalation, the image reconstruction with data
acquired during exhalation and inhalation is enabled. This means
that with the one cluster having the maximum measured respiratory
position of a first respiratory phase, a second respiratory phase
is regarded as adjacent to the cluster having the maximum measured
respiratory position, although the two clusters are assigned to
different respiratory phases. The characteristics of this
alternative are:
[0170] a) As before, a cluster is characterized by a respiratory
position range (or acceptance range, respectively) and a
respiratory phase. It comprises all shot indices (block 913) that
have already been measured after a respiratory position in the
respiratory position range assigned to the cluster, and the
respiratory phase assigned to the cluster, have been measured.
[0171] b) Directly adjacent clusters are assigned the same
respiratory phase, with one exception: The expiratory cluster
having the maximum respiratory position and the inspiratory cluster
having the maximum respiratory position can be regarded as adjacent
as long as c) is complied with.
[0172] c) Clusters having the same respiratory phase are arrayed in
accordance with their respiratory position range. The respiratory
position ranges of different clusters having the same respiratory
phase are disjunctive and "gap-less." The two adjacent clusters
having different respiratory phases have either the same
respiratory position range, or the respiratory position range of
the inspiratory cluster is connected, without gaps, to the
respiratory position range of the expiratory cluster, until it
reaches smaller respiratory positions. Both rules require, in
general, the insertion of empty clusters (i.e. of clusters having
an acceptance range/respiratory phase during which no data have yet
been acquired).
[0173] It should be noted here, that the neighborhood at the
transition between expiratory and inspiratory clusters is normally
temporary. If a more extreme respiratory position is measured, the
existing neighborhood relation between expiratory and inspiratory
clusters, each having a maximum respiratory position, is
interrupted, and further clusters are inserted.
[0174] d) As soon as an arbitrary group of n adjacent clusters
comprises a set of all of the shot indices [0, . . . , Ns-1] that
are to be measured, the measurement can be stopped (block 917),
because the overall variation to the respiratory position is thus
limited to the acceptance window.
[0175] The characteristics a)-d) are sufficient for an automatic
location of the final acceptance window. With the characteristic e,
a phase ordering can also be obtained.
[0176] e) Each cluster is assigned a starting position as the shot
index (also referred to as "start index ns_seed") in the interval
[ns0, . . . , Ns-1]. The cluster grows starting from the starting
position, in such a manner that the index range covered by the
cluster is without gaps. Optimally, thereby, the index range is
continued at the range borders in a periodic or cyclical manner (as
described above in reference to the optimized 2-bin PAWS). The
decision, whether a cluster grows toward smaller or larger shot
indices, also includes, in general, the shot indices already
acquired from adjacent clusters. The aim is to select an index in
such a manner that an arbitrary group of n adjacent clusters, which
contain the cluster assigned to the current measured respiratory
position, is as complete as possible, i.e. comprises as many
different shot indices as possible. The optimal distribution of the
starting positions between adjacent clusters depends on the number
of bins.
[0177] If a specific pattern of starting positions is obtained,
than this should also be maintained in the transitions between
expiratory and inspiratory clusters. This may require the insertion
of empty clusters.
[0178] In another embodiment example, a diminishing variance
algorithm (DVA) is selected as the determining algorithm for query
909. In his case, a possible expansion of the DVA can be
implemented, taking into consideration the respiratory phase, in
the following manner:
[0179] Respiratory positions that have been measured during the
respiratory phase "exhalation" and the respiratory phase
"inhalation" are recorded in separate histograms. At the end of the
initial phase (i.e. after all desired k-space data are recorded
completely, without gating, together with the respective measured
respiratory position and respiratory phase), the most frequent
respiratory phase is first determined, and the mode for each of the
two histograms is determined. In the subsequent re-acquisition
phase, first such k-space data that have been acquired during the
"unknown" phase are re-acquired. With each new measurement, the
histograms are updated (and thereby, the modes, respectively), and
the current most frequent respiratory phase is determined. As soon
as there are no more k-space data having the respiratory phase
"unknown," the mode of the histogram having the most entries
overall is selected as the basis for the re-acquisition (block
913). As is the case in the prior art, the k-space data having
respiratory positions deviating to the greatest extent from this
mode are taken into consideration for the re-acquisition. Due to
the digitalization of the respiratory signal, there are normally
numerous k-space data packets ("shots" in the notation used above),
having the same respiratory position. Among the shots that deviate
to the greatest degree from the selected mode, there may be some
that have been measured in different respiratory phases. If this is
the case, then (in deviating from the prior art) the repetition of
the shots is initiated, of those that were measured during the less
frequent respiratory phase. The termination criterion is either a
time limit and/or that all of the measurement data have been
measured during one respiratory phase, and the respiratory
positions lie within an acceptance window of a given width (block
917). It can be seen that, due to the continuous updating of the
histogram, the most frequent respiratory phase can, at least in
theory, change during the runtime.
[0180] FIG. 10 shows, in an exemplary fashion, an embodiment
example of a method described in reference to FIG. 9, in
conjunction with the new optimized 2-bin PAWS. For this, the
cluster diagram of a new 2-bin PAWS measurement is shown, with the
given differentiation of the respiratory phases. Each horizontal
bar corresponds to a cluster. These are arranged vertically with
respect to their respiratory position range. The shot index ns is
recorded in the horizontal axis. The respiratory position AP is
recorded along the vertical axis. A cluster contains, respectively,
the shots acquired from said cluster, along the horizontal axis.
The left-hand diagram shows the cluster having the respiratory
phase "exhalation" (E), and the right-hand diagram shows the
cluster having the respiratory phase "inhalation" (I).
[0181] In this case, the diaphragm position measured with a
navigator serves as the respiratory signal. The resolution was 0.5
mm. Accordingly, the respiratory signal is digitalized into 0.5 mm
units. Because only relative positions are measured with the
navigator, in relation to a reference state, the zero point of the
vertical axis is selected such that "0" corresponds to the maximum
measured end-expiration signal.
[0182] The width of the final acceptance window is set by the user
at .+-.1 mm. Due to the digitalization to a 0.5 mm grid pattern, it
therefore comprises 5 different respiratory positions. These are
distributed to peripheral and central clusters such that the
acceptance range of a peripheral cluster comprises three different
respiratory positions, and the one central cluster comprises
two.
[0183] The two final clusters are indicated with thinner lines. The
respiratory phase of these two clusters is "exhalation" (E). The
acceptance range of the central final cluster contains the
respiratory positions at -4.5 mm and -4.0 mm. The acceptance range
of the peripheral final clusters contains the respiratory positions
at -3.5 mm, -3.0 mm, and -2.5 mm.
[0184] One sees in FIG. 10 that only relatively few signal points
are excluded by the differentiation of the respiratory phases. The
total number of excluded shots is equal to the number of shots
assigned to the two encircled clusters having the respiratory phase
"inhalation" (I), the acceptance range of which corresponds in each
case to that of a final cluster.
[0185] A measured respiratory signal is depicted in an exemplary
manner in FIG. 11, as occurred in the measurement from FIG. 10. For
this purpose, the time t is recorded at the left, and the
respiratory position is recorded at the top. The measured
respiratory positions are recorded as circles or triangles,
depending on whether the respective respiratory position is
assigned to the phase "inhalation" (circles) or "exhalation"
(triangles). The acceptance window AF is indicated at the left edge
of the diagram, and by continuous lines. Furthermore, the overall
variation V1 and the overall variation V2 to the respiratory
position during the acquisition of the final measurement data sets,
which result when one undertakes a differentiation of the
respiratory phase (V1) or does not undertake such differentiation
of the respiratory phase (V2), is indicated with punctiform lines.
As one sees, the overall variation V2, in which adjacent
respiratory positions can lie in adjacent clusters, independently
of their respiratory phase, is much larger than when, as is the
case here, only respiratory positions having the same respiratory
phase can lie in adjacent clusters.
[0186] The imaging sequence here concerns a 3D double echo-played
gradient echo sequence. This acquires, directly after a navigator
sequence, in each case, 35 k-space lines having different values
for the two phase encoding gradients. Directly after 35 k-space
lines, another navigator sequence is carried out. An inherent
property of the prospective gating is that the decision process can
only be based on previous navigator measurements. However, use the
discrepancy between previous and subsequent navigator measurements
can be retrospectively used as the measure for the actual variation
to the respiratory position during the imaging measurement.
[0187] If the prospective decision is based exclusively on the
respiratory position, as is the case with the prior art, then the
actual variation to the respiratory position is V2 while the image
data used for the reconstruction are significantly larger than the
acceptance window AF. For the three respiratory cycles shown, this
lies between the punctiform lines with a spacing of V2. For the
overall measurement, it is even larger.
[0188] The four circled signal points in FIG. 11 have a respiratory
position, which lies within the acceptance window, but are not used
in the presented method for the final image reconstruction, because
at this point, the data measured directly after these signal points
are assigned to other clusters. The exclusion of these signal
points leads to a significant reduction to the actual total
variation to the respiratory position. For the three illustrated
respiratory cycles, it lies between the broken lines (spacing V1)
and is only slightly larger than the acceptance window.
[0189] By means of the detection and differentiation of the
respiratory phase used here, a significant reduction of the actual
respiratory movement during the imaging measurement and the thereby
associated respiratory artifacts is thus obtained through a
relatively small reduction in efficiency.
[0190] In the prior art, one possibility for reducing the
respiratory movement during the imaging measurement is to reduce
the size of the acceptance window. In order to obtain a
corresponding reduction to the actual respiratory movement of ca.
50%, as is the case in FIG. 11, a reduction to the size of the
acceptance window of likewise ca. 50% would also be necessary,
which would result in a significant reduction to the
efficiency.
[0191] One further possibility in the prior art would be to use
double gating methods, known from the article cited above, by Sachs
et al. "Real-Time Motion Detection in spiral MRI Using Navigators,"
MRM 32: Pages 639-645 (1994). This, however, is only compatible
with the simple acceptance-rejection algorithm. With the use of
this method, the disadvantages specified above associated with the
acceptance-rejection algorithm (efficiency loss with
varying/drifting respiration) must be accepted.
[0192] Although modifications and changes may be suggested by those
skilled in the art, it is the intention of the inventor to embody
within the patent warranted hereon all changes and modifications as
reasonably and properly come within the scope of his contribution
to the art.
* * * * *